W3C home > Mailing lists > Public > public-test-infra@w3.org > July to September 2019

Re: [css-fonts] Test for rem/em dimensions based on :root font-size

From: Philip Jägenstedt <foolip@google.com>
Date: Tue, 10 Sep 2019 13:04:37 +0200
Message-ID: <CAARdPYfTp=bqsG6wO+tjQQe+N6ry08D=ahdvuQ4dC2eVjK4dSA@mail.gmail.com>
To: Jon Ronnenberg <jon.ronnenberg@gmail.com>
Cc: public-test-infra <public-test-infra@w3.org>
If you look at https://github.com/web-platform-tests/wpt/pull/18941 and the
checks, you'll see "wpt.fyi - firefox[experimental]" and similar. You can
follow that to
and from there you can find a link that says "Latest results for PR
#18941", leading you to
https://wpt.fyi/results/?label=pr_head&max-count=1&pr=18941. There, it
appears Chrome has failed and Firefox and Safari have passed.

Note that the reason you don't see a "wpt.fyi - chrome[experimental]" check
is becaue https://tools.taskcluster.net/groups/Hfy0B5SiQ52jJxcI3031eg had a
failing task due to an Ubuntu mirror issue. When that sort of thing
happens, you can push more commits to trigger new runs, or just close and
reopen the PR.

On Tue, Sep 10, 2019 at 11:53 AM Jon Ronnenberg <jon.ronnenberg@gmail.com>

> We actually know why Chromium is behaving how it is.
> https://bugs.chromium.org/p/chromium/issues/detail?id=308862#c25 and the
> fix is https://bugs.chromium.org/p/chromium/issues/detail?id=308862#c32 (
> https://chromium-review.googlesource.com/c/chromium/src/+/1584288).
> On Tue, Sep 10, 2019 at 11:43 AM Jon Ronnenberg <jon.ronnenberg@gmail.com>
> wrote:
>> Thank you for you nice welcome.
>> It is really daunting to create the first PR, since the project is huge
>> in scope, has a lot bots from various issue trackers. I was afraid do ruin
>> someones build system or something dramatic. But it turns out that, even
>> though https://web-platform-tests.org/ is also huge in scope, the parts
>> I needed was actually written very concise. And  I created
>> https://github.com/web-platform-tests/wpt/pull/18941 which I eager to
>> get feedback on.
>> > I tried rem-height.html from the bug, but it actually seems to give the
>> same result in Chrome and Firefox on Linux.
>> > I've also tested percentage.html and still get the same behavior in
>> Chrome and Firefox.
>> Odd. rem-height.html only show difference in rem unit length when the
>> viewport is small enough - below 600px width. But percentage.html should
>> always show a larger square in Chrome than in Firefox, when click the
>> second button (30% root font-size). I just tested in Chrome Canary on
>> Windows.
>> https://github.com/web-platform-tests/wpt/pull/18941 has the test in
>> percentage-rem-low.html and a mismatch in percentage-rem-low-ref.html -
>> this is as good as a screen shot.
>> Is it possible that you can trigger CI screenshot of PR 1894?
>> Thank you for taking your time
>> On Tue, Sep 10, 2019 at 11:27 AM Philip Jägenstedt <foolip@google.com>
>> wrote:
>>> I added your test to
>>> https://github.com/web-platform-tests/wpt/pull/18957 just to see what
>>> the screenshots would be, and I've linked them there. Looks like the test
>>> would have to be made smaller to fit into 800x600. Not sure what the key
>>> difference you'd like to isolate is, but iterating on a PR like this in
>>> web-platform-tests should allow you to arrive at a useful test.
>>> On Tue, Sep 10, 2019 at 11:08 AM Philip Jägenstedt <foolip@google.com>
>>> wrote:
>>>> I've also tested percentage.html and still get the same behavior in
>>>> Chrome and Firefox.
>>>> On Tue, Sep 10, 2019 at 11:05 AM Philip Jägenstedt <foolip@google.com>
>>>> wrote:
>>>>> Hi Jon,
>>>>> I see that https://crbug.com/308862 has a lot of stars, so this is
>>>>> probably an important issue to resolve. Thanks for pushing it!
>>>>> Tests for parts of the web platform not yet tested are always
>>>>> appreciated! You've already found our main documentation site
>>>>> https://web-platform-tests.org/ and the ambition is that this should
>>>>> be enough to get you through the process. However, the repo is a little
>>>>> daunting with a lot of open issues and PRs, so having someone who you can
>>>>> poke for reviews and help, well, helps.
>>>>> A reftest is probably the right tool for this if it depends on a fixed
>>>>> viewport width, as reftests are always run at 800x600. Given a page that
>>>>> results in different rendering in different browsers it should be fairly
>>>>> straightforward to turn it into a reftest.
>>>>> I tried rem-height.html from the bug, but it actually seems to give
>>>>> the same result in Chrome and Firefox on Linux. Can you provide screenshots
>>>>> of that or another test case showing the differences between browsers?
>>>>> On Sun, Sep 8, 2019 at 11:17 PM Jon Ronnenberg <
>>>>> jon.ronnenberg@gmail.com> wrote:
>>>>>> 6 years ago I filed a bug in
>>>>>> https://bugs.chromium.org/p/chromium/issues/detail?id=308862 and
>>>>>> lately it has been picked up by both Chromium developers and web developers.
>>>>>> The short description is that, using percentage value in :root{
>>>>>> font-size }  below a certain threshold does not translate to correct rem/em
>>>>>> values in blink and webkit. It does work as advertised on w3, in Edge and
>>>>>> Firefox. <https://www.w3.org/TR/css-values-3/#rem
>>>>>> <https://www.w3..org/TR/css-values-3/#rem>>
>>>>>> I have look at the test cases in <
>>>>>> https://github.com/web-platform-tests/wpt/tree/master/css/css-values>
>>>>>> and searched in issues <
>>>>>> https://github.com/web-platform-tests/wpt/issues?utf8=%E2%9C%93&q=is%3Aissue+REM+font-size>
>>>>>> but have not found anything that suggest that there exist a test for this
>>>>>> issue.
>>>>>> Myself and others have provided several test cases in the chromium
>>>>>> issue < https://bugs.chromium.org/p/chromium/issues/detail?id=308862>
>>>>>> and I am wondering if it is feasible to translate a simple one of them into
>>>>>> a test in the web-platform-tests Project.
>>>>>> I don't have any experience writing these kind of tests and would
>>>>>> need help to get started. I imagine that it could be a reftest <
>>>>>> https://web-platform-tests.org/writing-tests/reftests.html> which
>>>>>> could be run automatically.
>>>>>> Is there interest in this kind of test, is it the right place for
>>>>>> such a test and how do I get started?
Received on Tuesday, 10 September 2019 11:05:14 UTC

This archive was generated by hypermail 2.4.0 : Friday, 17 January 2020 17:34:14 UTC